Application of the phase gradient autofocus algorithm to time delay estimation in linear arrays.

Abstract

For the purpose of localizing a distant noisy target, estimating the time delay of a wave front as it traverses across the sensors in a SONAR array is of considerable interest. This paper studies the application of the phase gradient autofocus (PGA) algorithm originally developed for extracting degrading phase errors in synthetic aperture radar (SAR) images to the time delay estimation problem in large linear arrays. It is well known that an efficient time delay estimation procedure consists of multiple cross correlations (one for each sensor pair) followed by a linear minimum-variance estimator. Monte Carlo simulation examples are presented that indicate the performance of the PGA method approaches the performance of the multiple correlator method at signal-to-noise ratios (SNRs) exceeding 0 dB with a substantial reduction in computational cost. Furthermore, PGA can easily take advantage of apriori knowledge regarding source bearings or array shape (if not perfectly linear) in order to improve the time delay estimates.